Scientists have tried over at least the past few hundred years (depending on your definitions) to build, from scratch, a perspective on the world which is as free from human bias as possible. At the moment, the jewel in the crown is quantum physics: an inherently statistical theory, so detached from human biases and assumptions that many smart people have struggled to understand or accept it, despite its incredible predictive power.<p>At the heart of the whole process is statistical inference: generalising the results of experiments or observations to the Universe as a whole. A "statistical crisis in science" would be terrible news. We may have been standing on the shoulders of the misinformed, rather than giants. Our "achievements", from particle accelerators to nukes and moon rockets, could have been flukes; if the underlying statistical approach of science was flawed, the predicted behaviour and safety margins of these devices could have been way off. We may be routinely bringing the world to the edge of catastrophe, if we don't understand the consequences of our actions.<p>Oh wait, it seems like some "political scientists" have noticed that their results tend to be influenced by external factors. I hope they realise the irony in their choice of examples:<p>> As a hypothetical example, suppose a researcher is interested in how Democrats and Republicans perform differently in a short mathematics test when it is expressed in two different contexts, involving either healthcare or the military.<p>The article criticises scientists' ability to navigate the statistical minefield of biases, probability estimates, modelling assumptions, etc. in a world of external, political factors like competitive funding, positive publication bias, etc. and they choose an example of <i>measuring how political factors affect people's math skills</i>!<p>To me, that seems the sociological equivalent of trying to measure the thermal expansion of a ruler by reading its markings. What do you know, it's still 30cm!